Fast Measurement of Dielectric Loss Angle With Time-Domain Quasi-Synchronous Algorithm

Fast Measurement of Dielectric Loss Angle With Time-Domain Quasi-Synchronous Algorithm
复制标题

DOI:
10.1109/tim.2014.2362839
复制
发表时间:
2015-04
影响因子:
5.6
通讯作者:
Kang Wang;Zhaosheng Teng;He Wen;Qiu Tang
Kang Wang;Zhaosheng Teng;He Wen;Qiu Tang
中科院分区:
工程技术2区
文献类型:
--
作者:
Kang Wang;Zhaosheng Teng;He Wen;Qiu Tang

文献摘要

相似文献

快速傅里叶变换(FFT)算法在异步采样和非整数周期截断下不可避免地会出现频谱泄漏和栅栏效应。因此基于FFT的介电损耗角(DLA)测量需要窗函数和谱线插值算法,需要大量的采样数据和较高的计算成本。因此,本文提出一种基于时域准同步采样技术的DLA测量方法,以实现快速测量和高精度。首先,利用时域牛顿插值算法准确估计出原始电压和电流采样信号的基频。然后,利用三次样条插值算法重构原始异步采样信号的准同步采样序列(QSSA)。最后根据等效电路模型和基于FFT的电压电流QSSA频域分析计算出DLA。本文提出的方法无需窗函数和谱线插值算法即可完美抑制谱泄漏和栅栏效应。仿真结果和嵌入式系统上的实现结果验证了本文算法的有效性,且采样数据较少,计算成本较低。
Fast Fourier transform (FFT) algorithm inevitably suffers from spectral leakage and picket-fence effect under asynchronous sampling and nonintegral period truncation. So the FFT-based dielectric loss angle (DLA) measurement needs window functions and spectral lines interpolation algorithm, which requires massive sampled data and high computational cost. Therefore, this paper proposes a DLA measurement method on the basis of time-domain quasi-synchronous sampling technique to achieve fast measurement and high accuracy. First, the fundamental frequencies of the original voltage and current sampled signals were estimated accurately with the time-domain Newton interpolation algorithm. Then, the quasi-synchronous sampled sequences (QSSA) of the original asynchronous sampled signals were reconstructed using cubic spline interpolation algorithm. Finally, the DLA was calculated according to the equivalent circuit model and frequency-domain analysis of voltage and current QSSA based on FFT. The proposed method in this paper can perfectly restrain spectral leakage and picket-fence effect without window functions and spectral lines interpolation algorithm. The results of simulation and the implementation on the embedded system have confirmed the effectiveness of the proposed algorithm in this paper with less sampled data and lower computational cost.